AI Developments: Robotaxis, Game-Inventing AIs, and OpenAI's Valuation Surge

Left alone with nothing to do, AI agents started playing

MIT's experiment with AI agents reveals their capacity for creativity and adaptation, raising questions about AI autonomy.

Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox

Hugging Face's new approach to database grading through RL rewards could enhance post-training agent performance.

OpenAI working on new $30B raise at $1.25T valuation ahead of IPO

OpenAI's potential $30B funding round could set a new record for private company valuations, indicating immense investor confidence.

Canva’s first AI image model plans designs before generating them

Canva's new AI image model, Canva Craft, introduces a planning phase that could revolutionize design generation.

Sure, let’s smoosh SpaceX and Tesla together, who cares

Speculation about a SpaceX and Tesla merger raises concerns over SpaceX's financial stability amid high borrowing.

Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?

The discussion on whether OpenAI should start its own companies highlights the tension between product sales and innovation.

Most Americans Say AI Is Bad for Society, Oppose More Data Centers

Public sentiment against AI's societal impact suggests a growing skepticism towards AI technologies and infrastructure.

Meta turns Muse sharing back on

Meta's reactivation of Muse sharing raises privacy concerns, as researchers highlight vulnerabilities in public links.

Meta’s AI could still scrape news sites, legal experts say

Despite changes to its terms, Meta's ability to train AI on news sites remains legally contentious, signaling ongoing industry challenges.